Discovery of food identity markers by metabolomics and machine learning technology

Abstract

Verification of food authenticity establishes consumer trust in food ingredients and components of processed food. Next to genetic or protein markers, chemicals are unique identifiers of food components. Non-targeted metabolomics is ideally suited to screen food markers when coupled to efficient data analysis. This study explored feasibility of random forest (RF) machine learning, specifically its inherent feature extraction for non-targeted metabolic marker discovery. The distinction of chia, linseed, and sesame that have gained attention as “superfoods” served as test case. Chemical fractions of non-processed seeds and of wheat cookies with seed ingredients were profiled. RF technology classified original seeds unambiguously but appeared overdesigned for material with unique secondary metabolites, like sesamol or rosmarinic acid in the Lamiaceae, chia. Most unique metabolites were diluted or lost during cookie production but RF technology classified the presence of the seed ingredients in cookies with 6.7% overall error and revealed food processing markers, like 4-hydroxybenzaldehyde for chia and succinic acid monomethylester for linseed additions. RF based feature extraction was adequate for difficult classifications but marker selection should not be without human supervision. Combination with alternative data analysis technologies is advised and further testing of a wide range of seeds and food processing methods.

Comparative Metabolomics and Molecular Phylogenetics of Melon (Cucumis melo, Cucurbitaceae) Biodiversity

“…To select molecular features that were relevant for the metabolic classification of melon accessions, we applied random forest (RF) machine learning technology. RF technology tuned towards metabolic feature selection enables the selection of small sets of molecular features that, if manually supervised, can be relevant for sample classification [29]. For this purpose, we used the set of 605 provisionally annotated molecular features from the combined >80,000 data set (Table S4).…”

“…To select annotated metabolic features that are relevant to classifying the C. melo accessions, random forest (RF) analysis was performed using the subset of annotated molecular features [27] from the combined data set (Table S4). Feature selection by RF technology was as described earlier [29] using hyperparameter optimization proposed by [75]. The R-package randomForest v 4.6-14 [76,77] was used for training classification trees.…”

Dysregulation of amino acids and lipids metabolism in schizophrenia with violence

“…The support vector machine (SVM) is a machine learning classifier and can mitigate the effects of noisy data [50]. Both have been widely been employed for biomarker discovery [51,52]. RF and SVM analyses were separately used to evaluate the importance of the differential plasma metabolites identified by the Mann-Whitney-Wilcoxon rank sum test and to select plasma metabolic biomarkers.…”

Cysteine and Methionine Biosynthetic Enzymes Have Distinct Effects on Seed Nutritional Quality and on Molecular Phenotypes Associated With Accumulation of a Methionine-Rich Seed Storage Protein in Rice

“…Metabolite profiling of milkripe seed tissue by gas chromatography–mass spectrometry (GC–MS) was performed essentially as previously described ( Erban et al, 2007 ; Erban et al, 2019 ). To enrich for polar primary metabolites and small secondary products, 60 mg samples were extracted with methanol and chloroform as described above for free methionine determination but with the following changes for GC–MS analysis: extraction volumes were scaled to the larger initial sample mass, 13 C 6 -sorbitol at 17 mg/L was added to the methanol extraction step to allow later correction of analytical variance, and 160 μl, rather than 100 μl, polar phase aliquots were taken for analysis.…”